Saxitoxin (STX) is one of the most potent neurotoxins on Earth. It is produced by marine and freshwater harmful algal blooms (HABs) and is the archetype of the paralytic shellfish toxin (PST) family. STX blocks voltage-gated sodium (NaV) channels, causing paralysis and death. To date, there is no antidote against STX, and it is classified as a chemical weapon. As HABs events increase globally, STX poses serious threats to public health and commercial fishing due to its accumulation in seafood. Despite its lethality to most vertebrates, some frogs resist STX poisoning. This phenotype is thought to rely upon soluble ‘toxin sponge’ proteins called saxiphilins (Sxphs). Sxphs have a single high-affinity STX-binding pocket that is hypothesized to compete for STX-binding with the molecular target, providing a resistance mechanism distinct from classic ‘target site’ ion channel mutations. Here, we test this idea directly in vitro and in vivo. We show that the canonical Sxph, RcSxph, from the American bullfrog (Rana catesbeiana) rescues multiple human NaV isoforms from STX block using electrophysiology. Further, we confirm that high-affinity STX-binding is maintained across a range of temperatures using isothermal calorimetry experiments. We demonstrate that a single dose of RcSxph counteracts STX neurotoxicity and lethality in mice using paradigms where STX and Sxph are administered together or sequentially in either order. Importantly, we show that this function requires the high-affinity STX binding site, as mutants lose efficacy in vitro and in vivo. Our findings provide the first in vivo validation of the ‘toxin sponge’ resistance mechanism and show that Sxphs protect against STX poisoning, highlighting the potential to harness this protein class as antidotes for PSTs and other toxins.